Medical History, 1985, 29: 353-374
KOCH'S POSTULATES IN RELATION TO THE
WORK OF JACOB HENLE AND EDWIN KLEBS
by
K. CODELL CARTER*
Among the essential elements of a theory of disease is the set of criteria that must be
satisfied to establish that some specific factor is the cause of a particular disease. For
example, in the germ theory various nonequivalent criteria have been proposed for
establishing that specific micro-organisms cause particular diseases.' Perhaps the
most famous criteria are the conditions known as Koch's postulates.
The postulates are invariably mentioned even in superficial discussions of Robert
Koch's work. However, different authors state the postulates differently,2 and many
authors provide no evidence that they have read Koch; often the only justification for
stating the postulates in a particular way is a superficial secondary work.3 There is
also disagreement about whether or not Koch adopted the postulates from his
predecessors. Some commentators, assuming that Koch derived the postulates from
Jacob Henle, refer to the postulates as the Henle-Koch postulates4 or even as the
Henle postulates.5 Others deny that Henle significantly influenced Koch's thinking
about causal criteria.6 Given that the postulates have been so influential, it is
remarkable that there have been no serious attempts to resolve these issues.
This essay will first determine how Koch stated and used the postulates in some of
his early papers. We will then examine the research tradition from which Koch seems
to have derived the postulates, and argue that Edwin Klebs was a more important
*K. Codell Carter, PhD, Professor of Philosophy, Brigham Young University, Provo, Utah 84602, USA.
1 For a survey of several sets of criteria see Alfred S. Evans, 'Causation and disease: the Henle-Koch
postulates revisited', Yale J. Biol. Med., 1976, 49: 175-195.
' For example, contrast the following two versions: "(i) Isolate the organism suspected of causing the
disease. (ii) Grow the organism in laboratory cultures. (iii) Inoculate a healthy animal with the cultured
organism. See if the animal contracts the disease. (iv) If the animal contracts the disease, examine the
animal and re-isolate the organisms that caused the disease." James H. Otto and Albert Towle, Modern
biology, New York, Holt, Rinehart & Winston, 1977, p. 196. "(i) The organism must be found in every
case of the disease. (ii) The organism must not be found as an accidental or harmless parasite in other
diseases. (iii) The organism, after its isolation from the body and cultivation in pure culture, must
reproduce the same disease." Alex Sakula, 'Robert Koch (1843-1910): founder of the science of
bacteriology and discoverer of the tubercle bacillus', Br. J. Tuberc. Dis. Chest, 1979, 79: 391.
3 For example, in an essay entitled 'Dr. Koch's postulates', Lester S. King bases his version of the
postulates on Dorland's The American illustrated medical dictionary (J. Hist. Med., 1952, 7: 350-361,
pp. 351f.).
4For example, Evans, op. cit., note 1 above.
5"Henle's Postulates were redrawn in scientific terminology as Koch's Postulates." Phyllis A.
Richmond, 'The germ theory of disease', in Abraham M. Lilienfeld (editor), Times, places, and persons,
Baltimore, Md., Johns Hopkins University Press, 1980, p. 85.
6 For example, Raymond N. Doetsch, 'Henle and Koch's Postulates', ASM News, 1982, 48: 555f.
353
K. Codell Carter
source than was Jacob Henle. Finally, we will consider one way in which Koch's
postulates differ from both Klebs' and Henle's causal criteria. I will argue that
whereas Klebs' and Henle's criteria for disease causation are satisfied by causal
agents that are only causally sufficient, Koch's postulates stipulate that causes of
disease must also be causally necessary.
Much of my argument rests on the distinction between causal necessity and causal
sufficiency and, since these terms are frequently misused, I will explain how they are
to be understood in this paper. For our purposes, a phenomenon C is sufficient for a
phenomenon E if the occurrence of Censures the occurrence of E. A phenomenon C
is necessary for a phenomenon E if the nonoccurrence of C ensures the
nonoccurrence of E. It is obvious, but frequently overlooked, that if one wants to
bring about some state of affairs, or to explain something that has happened, one
seeks a sufficient cause, since by bringing that cause to bear one can be certain that
the desired effect will follow. On the other hand, if one wants to prevent or to
eliminate some state of affairs, or to explain why something did not happen, one
seeks a necessary cause, since by preventing that cause one can prevent the undesired
effect. Thus, the subtle change from sufficiency criteria to criteria for causal necessity
may reflect fundamental changes in the purposes and goals of medical discussions of
disease causation.
Over the course of his career, Koch employed different criteria for establishing
causality. The postulates appear in only a few of Koch's more than one hundred
published papers, and, with the exception of one 1890 publication,7 all the papers in
which Koch mentioned or explicitly used the postulates appeared between 1878 and
1884. There are major differences in Koch's own statements of the postulates. We
will begin by examining Koch's causal criteria in three groups of publications: (i) his
first two anthrax papers (1876 and 1881); (ii) his two 1878 papers on wound
infections; and (iii) certain papers published between 1882 and 1884.8
Koch's first important paper (1876) concerned the aetiology of anthrax.9 He
reported the first successful attempt to trace the life cycle of a bacillus, described
some technical innovations such as the suspended-drop method for culturing bacilli,
and argued that a specific bacillus, Bacillus anthracis, was the cause of anthrax. This
paper contains no discussion of criteria for proving disease causality. Yet Koch felt he
had proved that Bacillus anthracis was the cause of anthrax,10 and by examining his.
This paper was 'Ueber bakteriologische Forschung'; it is in J. Schwalbe (editor), Gesammelte Werke
von Robert Koch, Leipzig, Georg Thieme, 1912, vol. 1, pp. 650-660.
8 In his papers on cholera and in the paper on bacteriological research Koch used yet other versions of
the postulates. After 1890, he continued to presuppose causal criteria, but he did not use the postulates at
all. However, for our purposes it is unnecessary to examine these sources.
9'Die Aetiologie der Milzbrandkrankheit, begrundet auf die Entwicklungsgeschichte des Bacillus
Anthracis', Schwalbe op. cit., note 7 above, pp. 5-26.
"
Koch frequently made comments such as this: "Until now [ 1882] a proof [that some disease is due to
a micro-organism] has been completed only for anthrax. For many other human infectious diseases, for
example, recurrent fever, the infected wound diseases, leprosy, gonorrhoea, only the simultaneous
occurrence of parasites and a pathological process is known, but the causal connexion has not been
proven." 'Die Aetiologie der Tuberkulose', ibid, pp. 428-455, p. 444.
354
Koch's postulates in relation to the work of Henle and Klebs
arguments it is possible to identify the criteria that he presupposed.
In his 1876 anthrax paper, Koch provided no significant original evidence that
bacilli were either necessary or sufficient for natural anthrax. His discussion rested
almost exclusively on artificially induced anthrax in test animals. Koch mentioned
that he had often examined animals that died of natural anthrax (i.6)." However, of
those examinations he reported only that he found bacilli in the spleen of an anthrax
horse-the only horse he had examined (i.21). Koch cited earlier researchers who
identified bacilli in natural cases, but he also mentioned other investigators who did
not find them.
As Koch explained in the introduction to his 1876 paper, his study was based on
the work of Casimir Davaine. Davaine's results were generally regarded as the
strongest evidence that anthrax was caused by Bacillus anthracis. Davaine's
argument was an argument for causal necessity; he reported that he always found
minute rods in the blood of anthrax animals, and that inoculations with blood that did
not contain the rods would not cause anthrax."2 Koch did not have access to
Davaine's original papers (i.105); he knew Davaine's work from abstracts which
clearly presented Davaine's argument as an argument for causal necessity."3 In
harmony with the abstracts, Koch noted that " [Davaine] asserted that the rods were
bacteria and that anthrax could occur only if these rods from anthrax blood were
present" (i.5). As Koch here described it, Davaine's view was that the rods were
necessary for anthrax. One objection to Davaine's view was that anthrax sometimes
seemed to have been caused by inoculations with bacteria-free blood. If so, this
would have shown that the rods were not necessary. Koch answered this objection by
arguing that the inoculated blood must have contained bacilli spores (i. 18). Since
spores were difficult to detect, one could reasonably assume that earlier researchers
had overlooked them. Thus, by explaining earlier failures to find bacilli, Koch
defended Davaine's thesis that they were necessary for the disease.
Thus, the context of Koch's work suggests that he was mainly interested in proving
that the bacilli were necessary for anthrax. This is also how Koch himself stated his
argument. He reported that he repeatedly sought to generate anthrax by inoculations
with anthrax materials that contained no bacilli, with hay-infusion bacilli, and with "a
species of bacillus that developed spontaneously and that was very similar to Bacillus
anthracis" (i.18). Since these experiments failed, Koch concluded that a substance
could generate anthrax only if it contained anthrax bacilli or their spores. "Anthrax
substances, whether fresh, decayed, or dried, can only cause anthrax if they contain
Bacillus anthracis or its viable spores. This removes all doubt that Bacillus anthracis is
"
To reduce the footnotes I consistently refer to Koch's collected works, ibid., in this way by volume and
page.
"' Davaine's most important results were announced in a series of papers in Comptes rendus de
l'Academie des Sciences in 1863 and 1864.
13 For example, Friedrich Steudener pointed out that Davaine found anthrax rods in the blood of all the
animals that died of anthrax. He observed that "Davaine regarded these organisms as the actual cause of
anthrax. Through their vegetation they destroyed the blood in a process similar to fermentation. He
supported this view especially by inoculation experiments. Only blood that contained the bacilli could
convey anthrax." Friedrich Steudener, 'Ueber pflanzliche Organismen als Krankheitserreger', Samml.
klin Vortr., Innere medizin, 1872, 1: 38: 283-308. pp. 303f. This argument establishes causal necessity
and it is precisely like the argument Koch gave in his first anthrax paper. Koch knew Steudener's paper; he
cited it in his essay on the infected wound diseases.
355
K. Codell Carter
the actual cause and contagium of anthrax." (i.21f). In this passage, Koch infers that
bacilli cause anthrax from the observation that anthrax occurs only if injected
substances contain bacilli or spores. Thus, he is presupposing necessity as the
criterion for identifying the cause of anthrax. Here, and in other passages, Koch
clearly regards necessity as decisive in establishing causality (e.g., i.19).
By contrast, Koch seems not to have regarded sufficiency as equally significant.
First, he knew that the mere presence of anthrax bacilli in an animal did not ensure
that it would become diseased; ingesting anthrax bacilli did not invariably induce
anthrax (i.19), some inoculation procedures were unreliable (i.6), and even among
exposed susceptible animals, vulnerability depended on various factors (i.213). So
Koch could not claim that the bacilli alone were sufficient to cause anthrax. As we
will see, in later papers Koch adopted various causal criteria that were similar to but
weaker than strict sufficiency. However, in the early anthrax papers one finds no such
criteria. After some early failures, Koch developed inoculation procedures that
invariably induced fatal anthrax in certain test animals. But in describing his
procedure, he observed only that it was important because it provided a test for the
viability of bacilli cultures (i.7). Nowhere in the 1876 paper did he suggest that these
inoculations, which reliably killed test animals, provided direct evidence that the
bacilli were the cause of anthrax. The fatal inoculations certainly influenced Koch's
thinking, but as he himself stated his argument, he seems to have been most
influenced by the unsuccessful inoculations with anthrax substances that contained
neither bacilli nor spores. Thus, Koch seems not to have regarded sufficiency as
central in proving that the bacilli were the cause of anthrax.
This conclusion is supported by Koch's remarks in his second anthrax paper
(1881).14 He quoted the passage discussed above, in which he concluded that the
bacilli were the cause of anthrax because only substances containing them induced
anthrax (i.175). After reviewing some additional evidence, he concluded, "thus,
anthrax never occurs without viable anthrax bacilli or spores. In my opinion no more
conclusive proof can be given that anthrax bacilli are the true and only cause of
anthrax." (i. 184). Here, as in the 1876 paper, his conclusive proof of causality is that
bacilli or spores are necessary. Koch acknowledged that critics "demand that
inoculated bacilli be totally removed from any associated substances that may
contain dissolved disease materials." But, Koch responded, "this is impossible . . .. No
one can take seriously such an undertaking." Here he rejected as impossible the
demand to prove that inoculated bacilli were sufficient; he then insisted again that his
earlier arguments, which established necessity but not sufficiency, "prove as
conclusively as possible that bacilli are the disease material".
All of this suggests that in his initial work on anthrax Koch's main criterion for
proving causality was showing that inoculated substances were effective only if they
contained bacilli or spores, that is, showing that either bacilli or spores were
necessary for artificial cases of disease.
Koch published two papers on wound infections. He first summarized his work in a
lecture in the 1877 meetings of the Gesellschaft deutscher Naturforscher und
14 Koch, 'Zur Aetiologie des Milzbrandes', Schwalbe, op. cit., note 7 above, pp. 174-206.
356
Koch's postulates in relation to the work of Henle and Klebs
Aerzte.'6 He then published a long paper on the same subject.'6 These papers, which
were both published in 1878, are significantly different and we will consider them
separately.
Koch began the summary paper by observing that the regular discovery of
micro-organisms in infected wounds did not imply that these diseases "come about
only when organisms enter the body and proliferate there, in other words, that the
diseases are parasitic. This is because in many cases of unquestionably infectious
diseases either no organisms are found or too few are found to explain the symptoms
or the fatal termination of the disease." (i.58). Here again, as Koch described it, the
issue is whether the organisms are necessary. Koch explained that he conducted
inoculation experiments to appraise these objections to the parastic conception. He
reported that by using ordinary techniques he was able to identify organisms in only a
few test animals, and these results were, therefore, inconclusive. However, by using
new staining and illumination techniques, which he had developed, he could identify
micro-organisms in every case (i.58f).
In the summary paper, Koch's main goal seems to have been to establish that
micro-organisms were necessary for wound infections by explaining why earlier
researchers sometimes failed to find them and by developing reliable techniques for
identifying them. Koch discussed his inoculation experiments, but, as in the anthrax
papers, he seems not to have regarded them as central in establishing causality. He
reported that the inoculations were useful in providing diseased animals in which
organisms could be demonstrated; he also observed that the inoculations proved that
the wound diseases were indeed infectious. He did not cite these experiments in his
argument for causality. In the summary paper, therefore, as in the first anthrax
papers, Koch seems to have been principally concerned with establishing that
micro-organisms were necessary.
In his long paper on wound infections, for the first time, Koch explicitly mentioned
criteria for disease causality. These criteria are the first version of the postulates. He
wrote that a conclusive proof of the parasitic origin of some disease "would require
that we find parasitic organisms in all cases of the disease, that they are present in
such numbers and distribution. that the disease symptoms can be explained, and that
a morphologically distinguishable organism is identified for every different disease."
(i.75). Thus, Koch specified these three criteria:
1.1. The micro-organism must be exhibited in all cases of the disease.
I.2. The distribution of the micro-organism must correlate with and explain
the disease phenomena.
1.3. For each different disease, a morphologically distinguishable
micro-organism must be identified.
Koch mentioned the same criteria in another passage, and he identified 1. 1 and 1.2 in
a third passage as well (i.100, 72).
The first criterion requires that the micro-organism is necessary for the disease;
this is the condition that seems to be central in Koch's 1876 and 1881 anthrax papers.
I.2. requires no discussion in this context. However, 1.3 requires attention.
15 Koch, 'Neue Untersuchungen uber die Mikroorganismen bei infektiosen Wundkrankheiten', ibid.,
pp. 57-60.
16 Koch,
'Untersuchungen uber die Aetiologie der Wundinfektionskrankheiten', ibid., pp. 61-108.
357
K. Codell Carter
To avoid circularity, it must be possible to distinguish parasitic organisms in some
way other than by the disease they supposedly cause. But different diseases could be
caused by morphologically indistinguishable organisms."7 One might imagine, for
example, that the. only difference between two kinds of bacilli is that they produce
slightly different toxins which affect their hosts differently. Thus, we must broaden
I.3 to include more than morphological differences. This is compatible with Koch's
later thinking; in 1884, he wrote that "morphological characteristics are not
normally sufficient to distinguish bacteria" (ii.61), and in 1890, he insisted that every
possible characterisitic of different strains of organisms be considered before
identifying them as of the same species (i.652f). Thus, I.3 changes through the course
of Koch's career. Yet all the versions of I.3 contrast so sharply with some of Koch's
other criteria, which we will later examine, that we will ignore these changes and
simply restate 1.3 in such a way as to avoid the defect we have mentioned and to make
it compatible with Koch's later thinking:
I.4. The micro-organism must be distinguishable in some way from organisms
that are associated with other diseases.
We will take 1.4 as the correct statement of Koch's third criterion.
We must now determine how I.4. relates to the claim that a given organism is
sufficient for some disease. If an organism is strictly sufficient for a given disease,
whenever the organism is present the disease-must occur. Thus, the organism could
not occur nonpathogenically or as the cause of some other disease. But to prove that
the organism always leads to the given disease, the organism must be
distinguishable in some way from every other organism, that is, I.4 must be
satisfied. Thus, I.4 follows from strict sufficiency. On the other hand, an organism
might satisfy I.4 and still not be sufficient for any disease. There are two important
possible counterinstances: first, distinguishable organisms that are usually
pathogenic might sometimes occur nonpathogenically. Koch sometimes denied that
bacteria can ever be found in healthy tissues (i.71f); if true, this would exclude the
nonpathogenic occurrence of bacteria. At this point in his career, however, Koch had
no conclusive evidence regarding the possibility that ordinarily pathogenic
organisms could occur nonpathogenically;18 perhaps for this reason, his remarks in
this paper are inconclusive. Second, the disease agent might not be the organisms
themselves but a substance with which they are always associated in diseased
animals. Neither Koch's suspended-drop method nor passing cultures through
animal bodies could exclude this possibility because, in addition to bacteria, liquid
media could always contain trace substances from the original diseased animal. Here
again, I.4 could be true but the micro-organisms themselves still might not be
sufficient. In subsequent papers, Koch acknowledged both of these possibilities.
However, in the 1878 papers, his remarks are inconclusive; one cannot be sure
exactly how much of this Koch himself understood at the time. The important point is
that I.4 follows from but does not imply strict sufficiency; it is a weaker condition.
It will be useful to refer to 1.4, and to certain other criteria that Koch subsequently
introduced, as "weak sufficiency criteria"; this phrase, of course, reflects the logical
17 One of Koch's reviewers, C. Hueter, pointed this out in Dt. Z. Chir., 1879, 11: 389-399, p. 396.
18 Apparently, Koch first discovered the phenomenon of healthy carriers in about 1893 (ii.168f, 215,
920).
358
Koch's postulates in relation to the work of Henle and Klebs
relation between each of these criteria and the logically stronger principle of strict
sufficiency.
Koch's discussion of each infected wound disease is loosely organized around his
three criteria. For each disease, Koch identified a particular organism and showed
that its distribution in a diseased animal explains the pathological alterations, some
of the symptoms, and the death of the animal. To some extent this confirms 1.2 and
I.4. As in the summary version of-his paper, Kock also attributed earlier failures to
identify the organisms to technical inadequacies, and he suggested that his improved
procedures would reveal organisms in all cases (i.100f). This part of his argument,
exactly like the arguments in his anthrax papers, indirectly supports the claim that
micro-organisms are necessary by explaining some apparent counter-evidence. It is
obvious, however, that Koch's evidence 1.1 and I.4 is relatively weak. In this paper,
as in his anthrax papers, Koch's discussion rests almost exclusively on inoculation
experiments. He never argues that natural infections involve the organisms he
considers, and he can apply his results to human infections only with an argument by
analogy (i.61, 100). This may partially explain why, in contrast to what he wrote
about anthrax and tuberculosis, Koch never claimed that his work on infected
wounds conclusively established causality.
Koch announced the discovery of the tuberculosis bacillus in a paper delivered in
1882.19 This sensational discovery probably did as much as any single
accomplishment to establish the domination of the germ theory. Later in 1882, Koch
published a second paper on tuberculosis.20 Other papers appeared over the next two
years culminating in Koch's monumental 1884 paper on the aetiology of
tuberculosis.21 In these papers, he meticulously followed specific criteria for
establishing disease causality; these criteria were similar to but slightly different from
those he employed in the paper on infected wound diseases. Their centrality in
Koch's reasoning in his famous and eminently influential tuberculosis papers
ensured that they would profoundly influence subsequent aetiological thought.
In the first tuberculosis paper, Koch observed that "the goal of the investigation is
first to exhibit certain parasitic forms that are foreign to the body and that could
possibly be the cause of the disease." (i.429). As in earlier papers, Koch introduced
techniques that enabled him to identify the parasitic bacillus, and he explained
earlier failures to identify it (i.432f). Koch reported finding the bacillus in materials
from dozens of natural cases of human and animal tuberculosis and in approximately
two hundred animals that has been inoculated with tuberculous materials. He
concluded that "tubercle bacilli occur in all tuberculosis disorders of humans and
animals" (i.433). Up to this point, Koch has argued that the bacilli are necessary for
the disease. Next, he observed that the bacilli "are distinguishable from all other
micro-organisms". Thus the unique organisms associated with tuberculosis satisfy
I.4.22
19 Koch, op. cit., note 10 above.
s Koch, 'Ueber die Aetiologie der Tuberkulose', Schwalbe, op. cit., note 7 above, pp. 446-453.
21 Ibid., pp. 467-565.
22 If so, however, Koch's argument is defective. I.4 stipulated that each causal agent must be
distinguishable from organisms associated with other diseases. But what Koch called tuberculosis included
such different disorders as scrofula and phthisis, which, as he knew, were caused by the same organism. 1.4
359
K. Codell Carter
In both 1882 papers, Koch observed that establishing a regular coincidence
between a disease and a specific organism does not prove causality. Indeed, he noted,
even identifying the organism in the organs where the disease is known to originate is
not conclusive (i.433). Of various possible strategies for proving causality, Koch
proposed to use 'a method that began by isolating the suspected disease agent in pure
culture. One could then demonstrate that the agent was a living parasitic organism.
"It then remains to prove that the isolated parasite is really the cause of the disease.
To accomplish this, one must show that animals inoculated with the pure culture
contract the original disease. The inoculation must succeed not only sometimes but in
every attempt as is achieved in such infectious diseases as anthrax." (i.446f). In
substantial sections of both 1882 papers, Koch reported his attempts to inoculate
tuberculous materials and cultures. The effort that he devoted to this part of the
argument, together with his own explicit assertions, show clearly that Koch regarded
this as the most decisive step in identifying the cause of tuberculosis. After reporting
his inoculation experiments, Koch concluded "these facts, taken together, show that
the bacilli in tuberculous substances are not merely coincidental with tuberculosis,
but cause it. These bacilli are the real tuberculosis virus." (i.442).
In 1882, Koch also published a paper criticizing Pasteur's attempts to immunize
animals against anthrax.23 In this paper, Koch described his own method of studying
infectious disease, and he contrasted his method with Pasteur's. Koch claimed to
begin by examining "all the body parts that are altered by the disease to establish the
presence of the parasites, their distribution in the diseased organs, and their relation
to the body tissues." (i.208). However, this investigation merely provided an
orientation, after which one could "begin to demonstrate that the organisms are
pathogenic and that they are the cause of the disease. For this purpose they must be
cultured pure and, after they have thereby been entirely freed from all the parts of
the diseased body, they must be inoculated back into animals, preferably of the same
species as those in which the disease was originally observed." Koch mentioned
tuberculosis as a disease in which these criteria had been fully satisfied. Thus, in this
1882 anthrax paper, Koch endorsed the same criteria for causality as in the
tuberculosis papers that appeard in the same year. Yet, as we have seen, he did not
actually follow these steps in identifying the cause of anthrax; his 1876 and 1881
papers reveal a significantly different strategy.
In 1883, Koch published a response to criticisms of his work on tuberculosis.24 In
this paper, he did not explicitly discuss causal criteria, but he clearly held the same
conception of causation as that advanced in his 1882 tuberculosis and anthrax
papers.25
Koch's monumental 1884 paper, 'Die Aetiologie der Tuberkulose', contains his
could only be satisfied by characterizing tuberculosis aetiologically. Koch may have been sensitive to this
problem (cp. i. 508). He admits that the same situation obtained with respect to anthrax (i.470).
23 Koch, 'Ueber die Milzbrandimpfung', Schwalbe, op. cit., note 7 above, pp. 207-231.
4 Koch, 'Critische Besprechung der gegen die Bedeutung der Tuberkelbazillen gerichteten
Publikationen', ibid., pp. 454-466.
'5 This paper also contains suggestions that tubercule bacilli may be strictly sufficient. Koch observed,
for example, that while bacilli were found in "well over one thousand" tuberculosis cases, they "were
never found in numerous control investigations of healthy persons." (i.455).
360
Koch's postulates in relation to the work of Henle and Klebs
most complete discussion of causal criteria; his position was the same as in the earlier
tuberculosis papers.
First it is necessary to determine whether the diseased organs contain elements that are not
constituents of the body or composed of such constituents. If such alien structures can be exhibited,
it is necessary to determine whether they are organized and show signs of independent life. Such
signs include motility-which is often confused with molecular motion-growth, propagation, and
fructification. It is also necessary to consider the relation of such structures to their surroundings
and to nearby tissues, their distribution in the body, their occurrence in various states of the disease,
and so forth. Such considerations enable one to conclude that there is probably a causal connexion
between the structures and the disease. Facts gained in these ways can provide so much evidence
that only the most extreme sceptic would still object that the organisms may not be the cause, but
only a concomitant of the disease. Often this objection has a certain justice, and, therefore,
establishing the coincidence of the disease and the parasite is not conclusive. In addition, one
requires a direct proof that the parasite is the actual cause. This can only be achieved by completely
separating the parasites from the diseased organism and from all products of the disease that could
be causally significant. If the isolated parasites are then introduced into healthy animals they must
cause the disease with all its characteristics. (i.469f).M
Koch used anthrax to illustrate these conditions (i.470).
These criteria are the same as those in the earlier tuberculosis papers; they can be
summarized as follows:
T. 1. An alien structure must be exhibited in all cases of the disease.
T.2. The structure must be shown to be a living organism and must be
distinguishable from all other micro-organisms.
T.3. The distribution of micro-organisms must correlate with and explain the
disease phenomena.
T.4. The micro-organism must be cultivated outside the diseased animal and
isolated from all disease products which could be causally significant.
T.5. The pure isolated micro-organism must be inoculated into test animals
and these animals must then display the same symptoms as the original
diseased animal.
T.I and T.2 correspond to 1.1 and I.4; together they stipulate that the organism is
necessary for the disease and that different organisms can identified for different
diseases. T.3 is the same as 1.2.
In the 1882 and 1884 papers, Koch regarded T.4 and T.5 as most decisive. We
must determine how these criteria relate to 1.4. As we have seen, 1.4 does not
stipulate that the organism is sufficient for the disease in question; given I.4, it is
possible either that the organism is sometimes nonpathogenic, or that the cause is not
actually the organism itself but rather some substance with which the organism is
regularly associated in diseased bodies. In the 1881 anthrax paper, Koch insisted that
it would be impossible to exclude this second possibility (i. 184). However, by 1882,
Koch's use of solid culture media produced results that were more reliable and more
conclusive than those obtained from the liquid media previously in use. His new
techniques enabled him to do precisely what, one year earlier, he had rejected as
impossible-to isolate bacilli from every contaminating influence and to inoculate
material that, for all practical purposes, consisted of pure bacilli. Yet, even with his
improved technology, Koch could not prove that ordinarily pathogenic organisms
26 Koch gave a similar list of criteria at (i.5 17). Here, as at (i.455), he suggested that tubercle bacilli may
be strictly sufficient.
361
K. Codell Carter
were never nonpathogenic. Indeed, in the tuberculosis papers, he admitted that
tubercle bacilli could sometimes exist in a suitable host without causing the disease
(i.443f). Thus Koch still could not stipulate that the bacilli were sufficient for
tuberculosis.
Koch had found inoculation procedures that reliably produced anthrax,
tuberculosis, and the infected wound diseases. Thus, in each of these cases if the
organisms were introduced in the proper way, they were always pathogenic. Having
used solid media to determine that the organisms themselves were the causal agent,
this meant thatproperly inoculated organisms were strictly sufficient. T.5 generalizes
this condition; it stipulates that for each disease there must be an inoculation
procedure that will always produce the disease.
Koch knew that T.5 could be satisfied even by organisms that were not themselves
strictly sufficient for the disease they caused; in spite of the reliable inoculation
procedure for tuberculosis, the presence of tubercle bacilli was not sufficient for the
disease (i.443f). On the other hand, if the presence of some organism were strictly
sufficient for some disease, inoculating the organism would always produce the
disease. Thus, once again Koch has identified a causal criterion that is logically
weaker than strict sufficiency. T.5 and I.4 are both implied by strict sufficiency, but
they probably bear no logical relation to each other. It is imaginable that separate
cultures of indistinguishable organisms could reliably induce different diseases when
inoculated into animals. If so, of course, one would assume that there was some
difference between the cultures, but T.5 could be satisfied even if no difference could
be found. Thus, given T.5, it is possible for I.4 to be false. Similarly, given I.4, the
disease could still be caused by a substance other than the organisms themselves and,
therefore, there might be no way of inoculating pure bacilli to produce the disease.
Thus, given I.4, it is possible for T.5 to be false. Koch apparently regarded 1.4 and T.5
as logically independent; otherwise there would have been no reason to include I.4 in
his statement of T.2 while, at the same time, including T.5 as a separate condition.
But, while I.4 and T.5 may be independent, unlike I.4, T.5 excludes the possibility
that some factor other than the organism is the causal agent. In this sense, T.5 can be
regarded as stronger than I.4; this may explain why Koch favoured T.5.
In the 1882, 1883, and 1884 papers, for the first time, Koch gave significant
attention to natural cases of disease. Thus, the tuberculosis papers are less dependent
on inoculation tests than are the earlier papers. Nevertheless, in the tuberculosis
papers Koch insisted that T.4 and T.5 were most decisive. T. 1, T.2, and T.3, which
alone involved natural cases of disease and which he had earlier taken as conclusive,
he now believed to show only that the organism is probably the cause.27 Yet Koch
never abandoned T.1, T.2 and T.3. Each tuberculosis paper reports evidence
that the bacilli are necessary and can explain the disease phenomena.
'7 It is not entirely clear what Koch meant in saying that one criterion is most decisive. Presumably, each
criterion is necessary and they are jointly sufficient. Thus, no one criterion alone can prove causality. He
probably adopted this way of speaking from some of his contemporaries who also used T.4 and T.5 and
who seem to have regarded them as conclusive.
362
Koch's postulates in relation to the work of Henle and Klebs
II
By the middle of the nineteenth century, pathological anatomy had produced
several schools of thought which collectively dominated European medical research.
Rudolf Virchow was a leading contributor to one branch of this research tradition.28
In 1847, in the first issue of his famous Archiv, Virchow discussed the methods and
goals of what he called scientific medicine.29 He asked, "How can one decide with
certainty which of two coexistent phenomena is the cause and which the effect,
whether one of them is the cause at all instead of both being effects of a third cause, or
even whether both are effects of two entirely unrelated causes?"30 He observed that
such questions could be answered "partly by observation ... and partly by animal
experiments".31 Of these, "experiment is the final and highest court ... forexperiment
alone is equally accessible to the entire world of medicine, and experiment alone
shows the specific phenomenon in its dependency on specific conditions, for these
conditions are arranged by choice." Similar ideas can be found in the writings of
many of Virchow's contemporaries.
Virchow regarded diseases as characteristic physiological processes resulting from
altered or abnormal influences on the body.32 In his view, one important goal for
pathological anatomy was tracing the development of such morbid processes.33 Thus,
Virchow and his colleagues frequently applied his strategy for determining causation
in showing that a particular lesion was the cause of other morbid changes. In such an
investigation, one usually examined corpses which revealed different stages of the
disease process, and one sought to induce similar processes in animals. Virchow
believed that particular internal lesions could be caused by different external
influences, but his work focused on the internal morbid process and, as he admitted,
he seldom gave attention to the extemal influences which initiated such processes.34
Of course, Virchow's general strategy for establishing causality could also be used to
identify these external influences.
In a discussion published in 1840, Jacob Henle, who was also a pathological
anatomist, proposed a criterion for identifying external disease causes. This criterion
conformed perfectly with the general strategy that Virchow subsequently discussed.
Henle conjectured that many diseases may be caused by parasitic micro-organisms.35
26 For a discussion of Virchow's relation to the research tradition of pathological anatomy see L.J.
Rather, 'Rudolf Virchow's views on pathology, pathological anatomy, and cellular pathology', Arch Path,
1966, 82: 197-204.
'9 Rudolf Virchow, 'Ueber die Standpunkte in der wissenschaftlichen Medicin', Virchows Arch. path.
Anat. Physiol., 1847, 1: 3-19. This paper is translated in Lelland J. Rather, Disease, life, and man, Stanford
University Press, 1958, pp. 26-39.
30Ibid., p. 15.
3 Ibid., p. 17.
32 Virchow said things like this in many of his essays. For example, ibid., p. 3. There is a more complete
discussion in Rudolf Virchow, Gesammelte Abhandlungen zur wissenschaftlichen Medicin, Frankfurt,
Meidinger, 1856, pp. 36-53.
33 See, for example, Virchow's illustrations of how one determines causality in Virchow, op. cit., note 29
above, pp. 17f.
34 Virchow admitted this in response to criticism by Edwin Klebs. See Rudolf Virchow,
'Krankheitswesen und Krankheitsursachen', Virchows Arch. path. Anat. Physiol., 1880, 79: 1-19, pp. 9f.
35 Jacob Henle, Pathologische Untersuchungen, Berlin, A. Hirschwald, 1840, p. 43. George Rosen
translated the most relevant part of Henle's book, 'On miasmata and contagia', Bull. Hist. Med., 1938,6:
363
K. Codell Carter
However, he noted that even if one regularly found living organisms in contagious
fluids within diseased bodies, one could still not infer that the organisms were more
than harmless saprophytes. The contagion could still be the fluids themselves rather
than the organisms. According to Henle, "one could prove empirically that [the
organisms] were really effective only if one could isolate... the contagious
organisms from the contagious fluids, and then observe the powers of each
separately."36 Henle was sceptical about the possibility of carrying out such a proof,37
and apparently he never tried to do so. His research, like that of Virchow and the
other pathologists, focused on internal disease processes, and he gave little attention
to external factors, such as invading micro-organisms, in response to which diseases
might originate. Yet Henle's criterion rests on generating artificial disease processes
in animals and, therefore, it conforms to the general research strategy that Virchow
described. Henle's criterion is also significant since it anticipated T.4 and T.5.
Partly because pathological anatomy focused on internal processes rather than on
external causes, medical researchers in the nineteenth century gave little attention to
micro-organisms as possible causes of diseases. One exception was Karl Mayrhofer
who, in the early 1860s, published a series of papers arguing that certain organisms,
which he called vibrions, were the cause of puerperal fever.38 To prove that the
organisms were the cause, and not merely harmless concomitants of the disease,
Mayrhofer tried to isolate them from other disease products and to introduce them
into healthy rabbits.39 Given available technology, Mayrhofer came close to
satisfying Henle's criterion. Both Mayrhofer and Carl Braun who was the professor
of obstetrics under whom Mayrhofer did his work, cited Henle's conjecture that
various diseases could be due to micro-organisms.40 Thus, while Mayrhofer did not
specifically associate his experiments with the procedure that Henle discussed, he
may have been aware of Henle's remarks. In any case, Mayrhofer's work conformed
to the empirical standards of observation and experimentation by which his
contemporaries sought to establish causal relations.
In the late 1860s, several researchers reported identifying micro-organisms in
specific disease processes.4" After Mayrhofer, Edwin Klebs was among the first who
tried to prove experimentally that these organisms were causal agents. Klebs had
been Virchow's pupil, and during the 1860s, he published several papers that were
911-983. Henle also considered this possibility in his Handbuch der rationellen Pathologie, Berlin, A.
Hirschwald, 1851, vol. 2., pp. 459-480.
36 Ibid., p. 43.
Immediately after the sentence just quoted, Henle observed that there is no way in which such an
experiment could be performed.
36 Mayrhofer's most important work was 'Zur Frage nach der Aetiologie der Puerperalprocesse',
Mschr. Geburtsk. Frauenkr., 1865, 25: 112-134.
39 Karl Mayrhofer, 'Untersuchungen uber Aetiologie der Puerperalprocesse', Med. Jb., 1863, 19:
28-42.
4 Mayrhofer, op. cit., note 38 above, p. 115; Karl Braun, 'Kindbettfieber', in John Baptist Chiari, Karl
Braun, and Joseph Spaeth, Klinik der Geburtshilfe und Gynaekologie, Erlangen, Ferdinand Enke, 1855,
p.480.
41 For a review of the literature see Felix Victor Birch-Hirschfeld, 'Die neuern
pathologisch-anatomischen Untersuchungen ueber krankmachende Schmarotzerpilze', Schmidts Jb.
Med., 1872, 155: 97-109.
364
Koch's postulates in relation to the work of Henle and Klebs
entirely compatible with Virchow's research programme. In 1872, however, Klebs
began focusing on the external causal agents that Virchow had generally ignored.
Klebs' first major paper dealt with the pathology of gunshot wounds,42 and,
according to Koch, contained "the first attempt to demonstrate a causal connexion
between bacteria and the infected wound diseases" (i.65). Over the next few years,
Klebs published other papers attempting to identify bacteria as the external causal
agents responsible for many infectious diseases. He became progressively more
critical of pathological anatomy and especially of Virchow.
In his 1872 paper, Klebs observed that "tracing the invasion and the course of the
micro-organisms can make causality probable, but the crucial experiment is to isolate
the efficient cause and allow it to operate on the organism."43 In 1875, he observed
that if one could show that "inflammation and other reactive changes follow, step by
step, the spread of the schistomycetes, then it is logical to infer a causal relation
rather than a simple coincidence."" Klebs pointed out that experimental evidence
could support the same conclusion. To obtain such evidence, one must "isolate
substances from the body and use them to induce further cases of infection." Klebs
claimed to have followed both approaches and to have obtained mutually supporting
results. His papers and his procedures for establishing causality were widely known
and discussed.45
During the middle 1 870s, several researchers tried to prove that specific organisms
were causal agents by isolating and inoculating them. In a text that Koch frequently
cited in his 1878 papers on infected wounds, Felix Victor Birch-Hirschfeld discussed
two strategies for establishing causality. He observed that Klebs and others tried to
explain the pathological significance of micro-organisms by correlating the
advancing parasites with the sequence of morbid changes.46 According to
Birch-Hirschfeld, the recognition that this was inconclusive led to experiments in
which "bacteria were isolated in various ways from the liquid constituents of
infectious substances. One then compared the results of inoculating with isolated
bacteria and with the other liquid materials."47 These two strategies are exactly those
discussed in Klebs' papers. Birch-Hirschfeld's survey, as well as the primary
literature, shows that these strategies were common in medical research in the late
1870s. Those who used these strategies frequently mentioned Henle's conjecture
that micro-organisms could cause diseases;48 some also mentioned Mayrhofer or
42 Edwin Klebs, Beitrage zur pathologischen Anatomie der Schusswunden, Leipzig, Vogel, 1872.
4 Ibid., p. 105.
4Edwin Klebs, 'Beitrage zur Kenntniss der pathogenen Schistomyceten',Arch. Path. Pharm., 1875,3:
305-324, p. 321.
45 For a few of the many examples see, F. Steudener, 'Ueber pflanzliche Organismen als
Krankheitserreger', Samml. klin. Vortr., 1872, Innere Medicin no. 14, 38: 283-308, p. 300; Leopold
Landau, 'Zur Aetiologie der Wundkrankheiten', Arch. klin. Chir., 1874, 17: 527-554, p. 529;
Birch-Hirschfeld, op. cit., note 41 above, p. 98 (Birch-Hirschfeld called Klebs' work "epochmaking");
and Max Schuller, 'Experimental Beitrage zum Studium der septischen Infection', Dt. Z. Chir., 1876, 6:
113-190, p. 160.
46 Felix Victor Birch-Hirschfeld, Lehrbuch der pathologischen Anatomie, Leipzig, F.C.W Vogel, 1877,
pp. 233f.
47 Ibid., p. 236.
48 See, for example, Leopold Landau, 'Ueber puerperal Erkrankungen', Arch. Gynaek., 1874, 6:
147-188, p. 174.
365
K. Codell Carter
Klebs as the first to have used these strategies to obtain experimental evidence of
causality.49
In 1877, Klebs addressed a plenary session of the Gesellschaft deutscher
Naturforscher und Aerzte on the subject of 'The revolutions in medical opinions in
the last three decades'.50 He argued that recent advances in medicine resulted from
associating specific diseases with particular disease agents, and identified three
procedures as relevant in establishing disease causation: (i) anatomical
investigations of diseased organs; (ii) the isolation and cultivation of disease germs;
and (iii) the initiation of new cases of the same disease by conveying germs to healthy
animals.51 Klebs mentioned "two ways that can be followed to establish the
significance of organisms in the generation of infectious diseases. (i) If organisms are
identified that are well characterized and that are found exclusively in the given
disease process, anatomical evidence can be conclusive. (ii) If the form of the
organisms provides no certain point of departure, it can be decisive to convey the
disease by means of organisms that have been isolated and cultivated outside the
body."52 This lecture, which was reviewed in general medical literature and became
well known in German medical circles,53 contains many ideas that figured in Klebs'
earlier papers. As before, he regarded the two "ways" as independent and
comparably effective. In the 1877 lecture, however, the way that is to be used
depends on whether or not the organism in question is well characterized and absent
from other disease processes. This condition, which does not appear explicitly in
Klebs' earlier papers, is almost precisely 1.4. If this condition is met, one is to show
that the morbid alterations of the disease correspond to the distribution and
concentration of the organisms-a requirement similar to I.2.5 Obviously, it is only
possible to confirm I.2 if the organisms can be identified. Klebs' second way, which
can be used even if the organisms are not well characterized, is isolation and
inoculation. This way is precisely T.4 and T.5. In this paper, therefore, Klebs
identified Koch's two major weak sufficiency criteria; Klebs' two ways correspond to
the first two versions of Koch's postulates.
Klebs' procedures for establishing causality are also similar to Virchow and
Henle's strategies. He established causality by observation and by animal
experiments, and, in some papers, regarded experiments as most decisive. Klebs had
been Virchow's pupil, and it is possible that his procedure for establishing causality
was a conscious application of Virchow's general research programme. Moreover,
Klebs frequently mentioned Henle's conjecture that micro-organisms may cause
" Ibid., p. 529; Johannes Orth, 'Untersuchungen uber Puerperalfieber', Virchows Arch. path. Anat.
Physiol., 1873, 58: 473-460; Friedrich Sander, 'Die Bakterienfrage zu London und Berlin', Dt. med.
Wschr., 1877, 1: 8-10; David Haussmann, 'Untersuchungen und Versuche uber die Entstehung der
ubertragbaren Krankheiten des Wochenbettes', Btrg. Geburtsh. Gynaek., 1874-5, 3: 311-421, pp. 312,
351.
50 Edwin Klebs, 'Ueber die Umgestaltung der medicinischen Anschauungen in den letzten drei
Jahrzenhten', Versammlung deutscher Naturforscher und Aerzte in Munchen, Leipzig, Vogel, 1878.
a1 Ibid., p. 49.
52 Ibid., p. 51.
53 For example, Klebs' lecture was reviewed in the Berl. klin. Wschr, 1877, 14: 594.
54 The conditions are similar but not exactly the same. Koch stipulated that the advance of the
micro-organisms must explain the disease phenomena, Klebs required only that it correspond to the
development of the disease.
366
Koch's postulates in relation to the work of Henle and Klebs
diseases. In his 1877 paper, Klebs observed that Henle's Pathologische
Untersuchungen, in which Henle conjectured that micro-organisms could cause
disease and mentioned his criterion for proving causality, had not received adequate
attention.55 Klebs may also have known Henle's criterion for proving causality.
Of course, there were important differences between Klebs and the earlier
pathological anatomists. Klebs noted that "Henle recognized, as the causes of disease,
only universal physical and chemical influences, the life impulses, the same factors
that have often been identified by others who wrote before and after him. The
concept of a specific cause of a disease, which is absolutely destructive of life, is alien
to him as to most other pathologists."5" Klebs mentioned that this was true of
Virchow. Klebs criticized the earlier preoccupation with internal disease processes.
While Klebs' causal criteria may have been based, in part, on the writings of Virchow
and Henle, Klebs realized that his application of those criteria was quite unlike that
of his predecessors.
Viewed in this light, Klebs' procedures for establishing causality can be seen as
part of a broad tradition that extends back at least into the 1840s. This tradition
included Henle's conjecture that micro-organisms cause diseases, and it was clearly
compatible with, if not directly dictated by, the basic research strategy of
pathological anatomy. We have seen that Koch's weak sufficiency criteria were
identical to the criteria that emerged from this tradition; we will now examine the
evidence that this tradition was the source of Koch's sufficiency criteria.
Many writers have assumed that at least T.4 and T.5 came directly from Koch's
teacher, Jacob Henle. Whether or not Koch was aware of Henle's criterion, there are
reasons for doubting that Henle was Koch's primary source for these ideas. First,
while Koch admitted that he owed Henle a great debt of gratitude (i.3), he never
suggested that Henle had influenced his thinking about causal criteria or even his
general interest in bacteriology. Indeed, Koch wrote that he received no
encouragement to study bacteriology while he was a student at Gottingen;
"bacteriology did not exist at that time" (i.3). The recollections of Elie Metchnikoff,
who came to Gottingen to work with Henle just at the time Koch was graduating,
confirm Koch's remark. "When, in 1866, I worked under Henle, in G6ttingen, at a
time when there were serious investigations on the microscopic agents of infectious
disease he remained indifferent and .., at no time did the question of contagious
disease come up in his laboratory."57
Second, in his early work on anthrax Koch isolated relatively pure cultures of
anthrax bacilli and inoculated them into test animals. However, at that time, Koch
seems not to -have regarded this procedure as particularly significant in establishing
disease causality. The causal arguments in his early anthrax papers stressed necessity
rather than sufficiency. The first weak sufficiency criterion appeared in his long
paper on infected wounds, and Koch first used isolation and inoculation as a weak
sufficiency criterion in his 1882 tuberculosis papers. If one assumes Koch adopted
5Klebs, op. cit., note 50 above, p. 47.
56 Ibid., p. 46.
5 Quoted in Hubert A. Lechevalier and Morris Solotorovsky, Three centuries of microbiology, New
York, Dover Publications, 1974, p. 65.
367
K. Codell Carter
T.4 and T.5 directly from Henle while he was Henle's student in Gottingen, it is
difficult to explain why these or other weak sufficiency criteria were not central in the
causal arguments of his early papers.
When Koch began studying anthrax as a district physician in Wollstein, he was
virtually isolated from everyone else who was studying micro-organisms. In his first
anthrax paper, Koch admitted a lack of access to certain important publications
(i.13f), and he used reviews and abstracts rather than the original papers. The
postulates do not figure in the first anthrax papers or in the summary version of the
1878 paper on infected wounds that he delivered before the Gesellschaft deutscher
Naturforscher und Aerzte.
The Gesellschaft was the same society before which Klebs had presented hisfamous 1877 lecture, and had, in the 1878 meetings, again criticized Virchow's
research programme.58 Klebs and Koch were often together at these meetings.59
Shortly thereafter, Koch published his long paper on wound infections-the paper
containing the first version of the postulates-and he cited several of Klebs' essays
including all three of those discussed above (i.65f, 70), as well as one of the
procedures by which Klebs sought to identify disease agents (i. 70). Koch credited
Klebs with having been the first to attempt to prove that micro-organisms caused
infected wound diseases (i. 65). It may have been because of Klebs' influence that
Koch began to think explicitly about causal criteria at the time of the 1878 meetings.
In any case, Koch cited Klebs in several papers published in the early 1880s (i. 133,
158f, 183). In his famous tuberculosis papers of 1881 and 1884, he cited Klebs but
did not associate him with T.4 and T.5 (i.433, 437, 468, 525, 529). In Koch's less
famous and less comonly read second 1882 paper on tuberculosis, however, the
association was explicit. He mentioned that various methods had been used in
attempting to prove causality, but that "the best method, the method used by
everyone who has been seriously occupied with these investigations, was introduced
and refined by Klebs." (i. 446). He described this method as, first, producing
successive pure cultures to separate the parasite from all disease products and,
second, inoculating the isolated parasites into test animals. Koch then observed that
his own investigations of tuberculosis followed this procedure.
Given that Koch himself acknowledged that Klebs was the source of T.4 and T.5,
and given that I.4 as well as T.4 and T.5 were so similar to Klebs' two ways for
establishing causality, it seems likely that Klebs was the primary source for Koch's
weak sufficiency criteria. Koch was a consummate technician, and his experimental
work certainly satisfied Klebs' criteria more rigorously than Klebs himself, or anyone
else, had been able to do . However, this should not blind us to the possibility that
Koch did not originate the strategy that his own experimental work followed.
58 Edwin Klebs, 'Ueber Cellularpathologie und Infectionskrankheiten', Tageblattder 51 Versammlung
deutscher Naturforscher und Aerzte, Kassel, Baier und Lewalter, 1878, pp. 127-134.
5 Bruno Heymann, Robert Koch, Leipzig, Akademische Verlagsgesellschaft, 1932, p. 236.
368
Koch's postulates in relation to the work of Henle and Klebs
III
I have argued that Koch's weak sufficiency criteria were derived from Edwin Klebs
and that Klebs' views were part of a broad tradition of causal thinking that was firmly
grounded in the basic research programme of pathological anatomy. We must now
consider an important difference between Koch's postulates and the causal criteria
espoused by Klebs, Virchow, and Henle.
We have seen that Koch's early aetiological papers contain no sufficiency criteria
at all. Even in the papers in which he stated the postulates, he did not use one
particular sufficiency criterion; instead, we find a series of different criteria, none of
which is used in more than a few papers. Koch never suggested that any one of his
weak sufficiency criteria was itself adequate to prove causality. Each time he stated
such a criterion he also stipulated that the causal organism must also be necessary and
that it must explain the diseases phenomena. Even in the 1882 and 1884 papers, the
papers in which Koch described T.4 and T.5 as most decisive, he stated these other
conditions and gave evidence that they were satisfied. One gains the impression that
T.4 and T.5, the criteria usually emphasized by Koch's commentators, represent
almost a temporary anomaly in Koch's thinking-an anomaly occasioned by his
contract with Klebs. Considering his writings as a whole, Koch was more consistently
and, perhaps, more fundamentally committed to causal necessity than to any form of
sufficiency.60 In contrast to Koch, neither Henle, Virchow., Klebs, nor any of Koch's
contemporaries who used Klebs' criteria required that causal agents be necessary.
Apparently, they all regarded the weak sufficiency criteria alone as conclusive. We
will confirm these claims by reviewing the writings of various researchers.
In his only explicit consideration of causal criteria, Henle wrote that "one could
prove empirically that [the organisms] were really effective only if one could isolate
the contagious organisms from the contagious fluids, and then observe the powers of
each separately.""6 Nowhere in the context of this passage did Henle suggest that any
other conditions must be satisfied in demonstrating causality. Although he observed
that even if living organisms were regularly identified in diseased bodies, this would
not prove that the organisms were more than harmless saprophytes, there is no
reason to interpret this remark as stipulating that only organisms that are always
identified in a certain disease process can be causal agents.62 Indeed, the whole point
of the passage is that inducing disease by inoculating pure organisms conclusively
demonstrates that the organisms are the cause. This quotation is from a chapter
entitled 'Miasmata and Contagia'; while some passages in this chapter suggest that
"contagia" may be necessary for some diseases, the discussion, taken as whole, is far
from conclusive. From his discussion it is obvious that, in addition to miasmata and
contagia, Henle recognized both various physical and chemical properties of the air,
for example, unusual or variable pressure and temperature,63 and other so-called
60 This claim can also be supported by examining the papers that Koch wrote after 1890, the year in
which he last mentioned the postulates. From that time on, necessity criteria dominate his causal
reasoning.
61 Henle, op. cit., note 35 above.
62 Yet this is how Henle is sometimes interpreted. See, for example, Anton Mayr and Barbara Bibrack,
'Ursache-Wirkungs-Relationen bei Infektionskrankheiten',Zbl. Bakt. Hyg., 1974,226:168-183, p. 173.
63 Henle, op. cit., note 35 above, pp. 910-913.
369
K. Codell Carter
sporadic factors' as possible causes of such diseases as cholera, plague, or puerperal
fever. For example, he observed that "external and internal exanthemata can arise
from many other local and general causes, apart from an infective matter, and fever
can be added to these just as to the miasmatic-contagious exanthemata."65 This, of
course, was entirely compatible with orthodox medical thought in the 1840s. For
example, at that time, typical European obstetricians identified both contagia and
epidemic atmospheric influences as well as general deprivation, worry, shame,
attempted abortion, fear of death, dietary disorders, exposure to cold, local miasmas,
difficult delivery, and retention and decomposition of the placenta as possible causes
of puerperal fever 6-one of the diseases that Henle discussed in this section of
his work. In the context of the early 1840s, a discussion of contagia or even of
ways to prove that a particular contagion causes a specific disease certainly does not
show that the author regarded contagia as necessary even for diseases that were
generally recognized as contagious. To draw such an inference is to interpret Henle
in the light of subsequent developments. So far as I can determine, Henle never
stipulated unambiguously that causal agents must be necessary for the diseases they
cause, indeed, nothing Henle wrote provides an unambigous basis for any of Koch's
causal criteria other than T.4 and T.5. At the very least, therefore, it is a gross
exaggeration to speak of the Henle-Koch postulates.
Similar observations hold for Rudolf Virchow. As Klebs pointed out,67 and as
Virchow himself admitted,68 the pathologists were much less interested in external
causal agents than in internal disease processes. Virchow, too, was primarily
interested in the causal sequence of lesions that constituted a given disease process,
although he claimed to have given more emphasis than most pathologists to external
causes.69 However, Virchow's discussions of criteria for causation never suggested
that any causes, internal or external, must be necessary as well as sufficient, and his
few scattered remarks on external causes clearly suggest that he accepted the
orthodox view that most diseases could originate from a range of unrelated external
factors, no one of which was necessary.70 Thus, there is no evidence that Virchow
would have accepted Koch's first postulate.
In 1877, Julius Cohnheim, a prominent pathologist and one of Virchow's pupils,
4 Ibid., p. 919.
65 Ibid., p. 964. Henle made related remarks in other places, but he was not entirely consistent. Cf. pp.
955, 966.
66 See, for example, the causal discussions in Eduard Lumpe, 'Die Leistungen der neuesten Zeit in der
Gynaekologie', Z. k. k. Ges. Aerzte Wien, 1845, 1: 341-37 1; Eduard Martin, 'Ueber den gegenwartigen
Stand der Lehre vom Puerperalfieber und die nachste Aufgabe fur die fernerer Bearbeitung derselben', Z.
rat. Med., 1846, 5: 1-44; or Braun, op. cit., note 40 above.
67 Klebs, op. cit., note 50 above, p. 46, note 58 above, pp. 133f.
6 Virchow, op. cit., note 34 above.
69 Ibid.
70 For example, in an 1861 discussion of puerperal fever, Virchow insisted that it was essential to
distinguish cases of puerperal fever "which clearly bear an epidemic character" from cases "in which local
infection can be demonstrated". He also illustrated the importance of predisposition by this analogy: "Just
as a man who is overheated will contract facial erysipelas if he exposes his face to a draught, so too the
puerpera, overheated by the process of delivery, can become ill if her uterus is chilled." These remarks
were made in a discussion which took place at the thirty-sixth meeting of the Gesellschaft deutscher
Naturforscher und Aerzte. The discussion was reported in the Mschr. Gebursk. Frauenkr., 1861, 18:
278-380.
370
Koch's postulates in relation to the work of Henle and Klebs
published the first volume of a textbook on pathological anatomy.7" Cohnheim's
conceptions of disease and disease causation were very similar to Virchow's. In this
book, as in his other writings, Cohnheim emphasized the experimental method in
gaining an understanding of causation and of disease processes.72 Cohnheim was
unusually sensitive to the significance of research into the relation between
micro-organisms and disease. Indeed, he concluded that the recently gained
understanding of the parasitic nature of trichinosis should induce physicians "to seek
a similar explanation for the infectious diseases".73 However, Cohnheim's
discussions of aetiology are perfectly compatible with earlier notions. He observed
that aetiology is an unbounded domain including "cosmic physics, meterology and
geology no less than the social sciences [Socialwissenschaften ] and chemistry as well
as botany and zoology."74 He also observed that aetiological discussions in common
textbooks included everything "from temperament to beds, from air electricity to
fungi and fleas, from inheritance to drinks."75 Cohnheim's discussions of specific
diseases illustrate perfectly these general remarks. For example, as causes of "local
anaemia", he listed too high a resistance in the circulation, which can be caused by
low temperatures, stimulation of the vasomotor system by nerves or by certain
poisons such as opium or lead, as well as morbid changes in the artery walls from
calcification, reduction in elasticity of the veins, thrombosis, strictures, and abnormal
pressure.76 Thus, Cohnheim traced the specific abnormality to a variety of unrelated
external causes, no one of which was necessary for all cases of the disease.
As one would expect, Henle, Virchow, Cohnheim, and their colleagues focused
heavily on sufficiency criteria for disease causality. Here, for example, are the causal
criteria that appeared in an 1850 monograph on cholera: "In order for any purported
cause to be recognized as the actual cause of a disease, we must always be able to
answer the following questions affirmatively: does this cause always have the same
effect? As an experiment can one always bring about the disease in this way? In those
cases in which the cause does not bring about the specified disease, can the same
reason for failure always be identified?"77 Each of these conditions would be satisfied
by a cause that was sufficient or weakly sufficient, but none of them would be
satisfied by a cause that was necessary and not sufficient. Here, as a second example,
is an excerpt from an 1878 essay by Cohnheim entitled 'On the role of pathological
anatomy'. Cohnheim observed, "A really conclusive proof of a causal relation ...
can be given only when one succeeds, by use of the supposed cause, in generating at
will the desired effect, that is, the specific disease.... There can be no doubt about
the causal relation once one ... brings about the same alteration in rabbits or dogs
71 Julius Friedrich Cohnheim, Vorlesungen uber allgemeine Pathologie, Berlin, August Hirschwald,
1877. The second volume appeared in 1880.
72 For example, see Cohnheim's lecture 'Ueber die Aufgaben der pathologischen Anatomie', reprinted
in his Gesammelte Abhandlungen, Berlin, August Hirschwald, 1885, pp. 605-622.
73 Cohnheim, op. cit., note 71 above, pp. 251f.
Ibid., p. 8.
7 Ibid., p. 9.
76 Ibid., p. 80. Causes of "chronic nephritis" include "malaria, constitutional syphilis, chronic lead
poisoning, excessive consumption of alcohol, repeated exposure to cold, soaking wet conditions of part of
all the body, for example, of the feet" (vol. 2, p. 336).
77 Joseph Hamernik, Die Cholera epidemica, Prague, Calve, 1850, pp. 247f.
371
K. Codell Carter
that the [cause is supposed to have brought about in human cases of illness]."" The
aetiological literature of these decades contains many similar passages. Here again,
obviously, the issue is exclusively whether the supposed cause is sufficient. The
exclusive interest in sufficient rather than necessary causes also explains why
physicians in the period repeatedly warned against inferring a common cause from a
common effect.79
One observes precisely the same priority of interests in the writings of Edwin
Klebs. In the papers quoted above, Klebs seems clearly to have regarded isolation
and inoculation as conclusive. He never suggested that causal organisms must also be
necessary. The closest Klebs ever came to suggesting a necessity criterion is in the
following passage from his 1877 lecture: "In every adequately examined disease,
particular organisms are found that always function in the same way. This indicates
that these organisms stand in an important relation to the disease process. However,
that the organisms stand in a causal relation to the disease process follows from the
fact, conclusively established for at least some diseases, that conveying the organisms
always generates the same disease."80 The first sentence is ambiguous; however,
even if one interprets it to mean that causal organisms must be found in every case of
a given disease, the second sentence clearly shows that Klebs regarded the weak
sufficiency criterion alone as the decisive proof that the relation between the
organism and the disease is causal.8"
In his 1877 lecture, Klebs criticized Henle for failing to recognize the existence of
what Klebs called "specific causes". "The concept of a specific cause of a disease,
which is absolutely destructive of life, is entirely alien to [Henle] as to most other
pathologists."82 One may think that Klebs' use of the term "specific causes" provides
some evidence that he regarded causal agents as necessary. At least this term is now
often used in this sense.83 However, since Klebs never explicitly asserted that causes
of diseases must be necessary, it seems more plausible to interpret his use of this term
in a different way. As the context of this passage shows, Klebs was criticizing Henle
for thinking that the causes of diseases were simply ordinary processes occurring
78 Cohnheim, op. cit., note 72 above, pp. 615.
7 "Since it is in the nature of the nervous system to react in the same manner to the most varied stimuli,
to infer similar causes from similar effects as necessary as it otherwise is, is not entirely dependable."
(Henle, op. cit., note 35 above, p. 955); In a discussion of pathological changes in the body, Edward
William Murphy observed that "identity of effect by no means proves one and the same cause."
('Puerperal fever', Dublin Quart. J. Med. Sci., 1857, 24: 1-30, p. 4.)
80 Klebs, op. cit., note 50 above, p. 50.
8 We have seen that Klebs charged Virchow with ignoring external cause and focusing exclusively on
the internal causal sequence of the disease process. For his part, Virchow responded that by ignoring
internal disposition, Klebs was in danger of going to the opposite extreme. "Evidently Klebs is ... of the
opinion that the internal arrangement of the tissues is irrelevant [in determining the response to an
external cause], or, expressed in the terms of universal pathology, that the external cause is an Agens
Causa sufficiens for all the consequences." (Virchow, op. cit., note 34 above, p. 9). But, of course, Virchow
did not accuse Klebs of an interest in necessary causes.
8Klebs, op. cit., note 50 above, p. 46.
"By 'specific causal factor' is meant the enviromental factor or 'remote' cause-cause outside the
body-without which there could be no disease. Other environmental factors might or might not be present
but this factor was necessary for the disease to occur." Richmond, op. cit., note 5 above, p. 84. Lester S.
King says essentially the same thing in his discussion of disease causation in The philosophy of medicine,
Cambridge, Mass., Harvard University Press, 1978, pp. 209-213.
372
Koch's postulates in relation to the work of Henle and Klebs
under extreme or abnormal conditions, and for failing to recognize that causes of
disease could be specific in the sense of being totally unlike ordinary life
processes-the causes of diseases were specific to disease processes. This sense of
specific, which seems most suitable given the rest of Klebs' discussion, is perfectly
compatible with the possibility that the cause of a disease is sufficient or weakly
sufficient but not necessary.
Koch's postulates require that the cause of a given disease must be necessary and,
in one way or another, sufficient or weakly sufficient for the disease. On the other
hand, most of those who seem to have influenced Koch's causal thinking focused on
causes that were weakly sufficient but not necessary. I certainly do not wish to
suggest that no one in the 1870s and 1880s was interested in necessary causes of
disease. Indeed, in an earlier paper I examined a large body of literature that focused
precisely on necessary causes.84 Here I wish only to establish that, in contrast to
Koch, many medical writers in the middle of the nineteenth century gave primary
attention to causal sufficiency and often tended to ignore necessity. This seems to
have been particularly true of Henle, Virchow, Klebs, and of several other
researchers who contributed to the research tradition from which Koch's weak
sufficiency criteria were probably drawn.
As we observed above, one generally seeks a sufficient cause if one wants to bring
about some state of affairs or to explain why something has happened. Thus, we can
understand one factor that focused the attention of pathological anatomists on
sufficient causes; one of their objectives was to demonstrate causal relations by
inducing disease processes in animals. On the other hand, if one wants to prevent or
eliminate some state of affairs, one seeks a necessary cause. Given causes that are
sufficient but not necessary it is virtually impossible to undertake effective steps to
prevent or to eliminate an undesired effect. Thus, if one assumes that the goal of
medicine is to prevent or to eliminate disease, one would expect that physicians and
medical researchers would be primarily interested in necessary causes (whether or
not they are also sufficient). As we have seen, Koch always required that the
cause of a disease must be necessary, and he usually required that it must also
be sufficient or weakly sufficient. By contrast, the causes that Henle, Virchow, Klebs,
and several other researchers identified were sufficient or weakly sufficient but not
necessary. Thus, this rather subtle change in talk about disease causality may reflect
fundamental changes in the purposes and goals of medicine as an institution-changes
that are to be dated to about the middle of the nineteenth century.
In his 1877 and 1878 lectures, Edwin Klebs observed that while pathological
anatomy had made great strides in describing and explaining disease processes, it had
contributed almost nothing to the control of disease.85 Whether or not Klebs'
criticism was accurate, his remarks suggest that pathological anatomy had seriously
failed to satisfy earlier expectations. As Klebs himself clearly recognized, this failure
was partially a consequence of the pathologists' conceptions of disease causality.
8 K. Codell Carter, 'Ignaz Semmelweis, Carl Mayrhofer, and the rise of germ theory', Med. Hist., 1985,
29: 33-53.
85 Klebs, op. cit., note 50 above, p. 47, op. cit., note 58 above, pp. 133f.
373
K. Codell Carter
Klebs' own recommendation was that instead of concerning themselves with internal
causes of disease processes, physicians should investigate the external causes that
initiated each disease as a whole. Klebs, of course, believed that micro-organisms
were external causes that deserved primary attention. Unfortunately, his
recommendation was not adequate to solve the problem. He urged physicians to
think of a new range of causes, but his conception of the causes of disease was
essentially the same as that of the pathological anatomists, as sufficient causes of
particular events. In fact, what was required was not simply a new application of this
existing conception, but a subtle and yet profound change in the basic conception
itself. What was required was precisely what Koch's first postulate provided- the
concept of a necessary cause.
374